NEPG: Partitioning Large-Scale Power-Law Graphs

We propose Neighbor Expansion on power-law graph(NEPG), a distributed graph partitioning method based on a specific power-law graph that offers both good scalability and high partitioning quality. NEPG is based on a heuristic method, Neighbor Expansion, which constructs the different partitions and greedily expands from vertices selected randomly. NEPG improves the partitioning quality by selecting the vertices according to the properties of the power-law graph. We put forward theoretical proof that NEPG can reach the higher upper bound in partitioning quality. The empirical evaluation demonstrates that compared with the state-of-the-art distributed graph partitioning algorithms, NEPG significantly improved partitioning quality while reducing the graph construction time. The performance evaluation demonstrates that the time efficiency of the proposed method outperforms the existing algorithms.